Source-linked AI summary
Entanglement-Based Machine Learning on a Quantum Computer
X. -D. Cai, D. Wu, Z. -E. Su, M. -C. Chen, X. -L. Wang, L. Li, N. -L. Liu, Chao-Yang Lu, Jian-Wei Pan
TL;DR
Classical computers may struggle with machine learning on rapidly growing high-dimensional data, motivating quantum approaches. The paper experimentally uses entanglement-based distance evaluation on a small photonic quantum computer to classify 2-, 4-, and 8-dimensional vectors for supervised and unsupervised learning. The experiments demonstrate quantum manipulation and classification of high-dimensional vectors, while the scheme’s exponential speedup applies to vector dimension but not the number of training samples.
Problem
Machine learning on rapidly growing big data could become intractable for classical computers, motivating quantum algorithms for tasks involving large vectors.
Method
The paper uses entangled photonic quantum states and projective-measurement probabilities to evaluate vector distances and classify 2-, 4-, and 8-dimensional samples into two clusters.
Results
Two of 100 two-dimensional test samples were misclassified, while one of 17 four-dimensional and one of 9 eight-dimensional samples were misclassified experimentally.
Takeaways & Limitations
The experiments demonstrate that quantum computers can manipulate high-dimensional vectors and estimate distances and inner products used in machine-learning routines.
Takeaways & Limitations
The scheme can in principle achieve exponential speedup with vector dimension N, but not with the number of training samples M.
Abstract
from arXiv · showhide
Machine learning, a branch of artificial intelligence, learns from previous experience to optimize performance, which is ubiquitous in various fields such as computer sciences, financial analysis, robotics, and bioinformatics. A challenge is that machine learning with the rapidly growing "big data" could become intractable for classical computers. Recently, quantum machine learning algorithms [Lloyd, Mohseni, and Rebentrost, arXiv.1307.0411] was proposed which could offer an exponential speedup over classical algorithms. Here, we report the first experimental entanglement-based classification of 2-, 4-, and 8-dimensional vectors to different clusters using a small-scale photonic quantum computer, which is then used to implement supervised and unsupervised machine learning. The results demonstrate the working principle of using quantum computers to manipulate and classify high-dimensional vectors, the core mathematical routine in machine learning. The method can in principle be scaled to a larger number of qubits, and may provide a new route to accelerate machine learning.
Supplemental Material
Figure S1 demonstrates supervised nearest-neighbour classification by comparing testing vectors with reference vectors, then updating classifications when a new training vector is added.
- Eight testing vectors are classified by assigning each to the group of the closest reference vector, yielding A–D as blue and E–H as red.Distances to the two training vectors determine each label.
- Adding training vector R3 changes vector A’s label from red to blue because A is closer to R3 than R1.The other testing-vector labels remain unchanged.
- The classification boundary shifts after R3 is incorporated, changing from one theoretical boundary to two linked dashed lines.